+1 (201) 771-1665 Sales Support 24/7 Engagements active in 5 regions
Engagements

Rag Production Sprint.

Four weeks to a live retrieval-augmented generation system. Chunking, embedding, retrieval pipeline, and a queryable endpoint your team controls -- not a prototype.

Four weeks to a live retrieval-augmented generation system your team can query and monitor. Not a prototype. A running production deployment with an API endpoint, evaluation results, and full documentation.

What you receive

Not included

Frontend or chat UI. The sprint delivers the retrieval backend and API. UI integration is a separate scope item.

Four weeks to a running RAG deployment your team can query and monitor, not a prototype that stalls at demo stage.

Why teams choose Cobham for a RAG production sprint

How this works

Scoped before you sign.

Every engagement starts with a written brief that defines deliverables, duration, and a fixed price. No hourly billing. No vague statements of work. No surprise invoices when the scope runs over.

01

30-minute call

We map your current situation and tell you whether this engagement type fits the work in front of you. No obligation.

02

Written brief

A plain-language document: what we deliver, when, and at what fixed price. If the scope changes, we renegotiate before extra work begins.

03

Delivery & handoff

Runbooks, IaC, documentation, and full access credentials handed to your team at close. You own everything. No vendor lock.

Pricing

Flat rates. No markup. No surprises.

A

You own the cloud bill

Your cloud provider bills you directly at published prices. We never touch the invoice, never add a margin.

B

One flat rate

One number per engagement. The rate covers the full scope defined in the contract. No hourly overage.

C

30-day exit on retainers

Cancel any monthly retainer with 30 days written notice. Full handoff included at no extra cost.

Common questions

Frequently asked.

What document formats do you support for ingestion?
PDF, Word, Excel, plain text, HTML, and Markdown. For structured data (databases, APIs), we build custom ingestion connectors. Format support is confirmed during scoping.
Which vector store do you deploy?
Qdrant for most deployments -- mature, fast, and self-hostable. pgvector when the team already runs PostgreSQL and wants to minimize infrastructure components. Weaviate for workloads requiring hybrid search at scale.
What embedding model do you use?
Nomic Embed or BGE by default for local deployments. OpenAI text-embedding-3-small when API-based embedding is acceptable. The choice depends on your data residency requirements and volume.
How do you handle document updates after the sprint?
We document the re-indexing procedure and build a trigger mechanism (manual, scheduled, or webhook-driven) as part of the sprint. Your team can update the index without our involvement.
All engagement types
Rag Production Sprint

Get pricing in 30 minutes.

We publish the engagement types but not the rates -- every scope is different. Schedule a call and we will tell you exactly what this engagement costs before you sign anything.

Schedule a call
Direct to a principal · Response under four hours